- From: Doug Bunting <Doug.Bunting@Sun.COM>
- Date: Fri, 02 Jul 2004 11:42:24 -0700
- To: "Cutler, Roger (RogerCutler)" <RogerCutler@chevrontexaco.com>
- Cc: Sanjiva Weerawarana <sanjiva@watson.ibm.com>, Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>, www-ws-arch@w3.org
Roger, After a bit of asking around, I am pretty sure this exact technique is not mandated in any specification. However, as Sanjiva and others have indicated, support is pretty widespread. ?wsdl is a convention (but not a mandate) that helps to keep the simple things simple. Making complex things possible might require extensions to this convention, other conventions or new specifications. Other products supporting this useful convention include the Java Web Services Developer Pack[1] and Sun Java System Application Server[2]. For the Java platform in general, the JSR-109 (final release) specification[3] has jurisdiction over WSDL publishing at service deployment time. It says that deployed WSDL documents MUST be published to at least one location, where "location" means file, URL or registry. All application servers MUST support URL-based (or file-based) publication and they are encouraged to support registry-based publication too. The choice of URL / filename is entirely left to the deployer and the spec doesn't offer any guidelines, defaults or naming conventions in that area. The specification does, however, leave plenty of room for the specific convention you asked about. thanx, doug [1] http://java.sun.com/webservices/downloads/webservicespack.html [2] http://wwws.sun.com/software/products/appsrvr/home_appsrvr.html [3] http://www.jcp.org/en/jsr/detail?id=109 On 01-Jul-04 21:09, Sanjiva Weerawarana wrote: > MessageApache Axis supports the ?wsdl technique to get the WSDL for the > service. > > http://ws.apache.org/axis. > > Sanjiva. > > ----- Original Message ----- > From: Cutler, Roger (RogerCutler) > To: Cutler, Roger (RogerCutler) ; Savas Parastatidis ; www-ws-arch@w3.org > Sent: Friday, July 02, 2004 4:09 AM > Subject: RE: Requesting WSDL Files > > > Oh, wait a minute, that was dumb. Sorry. > > Microsoft has chosen, if enabled, to return a test-bed for the service from > the simple GET. Seems like a reasonable choice to me, but I don't really > want to debate that. What I'd really like to know is whether other vendors > are using the "?wsdl" convention. I have heard from one person who seems to > think that it is at least fairly widespread, but I don't have details. > > > > > From: Cutler, Roger (RogerCutler) > Sent: Thursday, July 01, 2004 5:02 PM > To: 'Savas Parastatidis'; www-ws-arch@w3.org > Subject: RE: Requesting WSDL Files > > > How do you distinguish, then, between a GET that is intended to return a > WSDL file and a GET of a Web Service that takes no parameters but returns > something? At least in the implementation I am familiar with, if GET is > enabled for a Web service that takes no parameters I think it's just the > base URL that invokes it. > > From: Savas Parastatidis [mailto:Savas.Parastatidis@newcastle.ac.uk] > Sent: Thursday, July 01, 2004 4:49 PM > To: Cutler, Roger (RogerCutler); www-ws-arch@w3.org > Subject: RE: Requesting WSDL Files > > > If we assume HTTP, I would prefer the even simple approach of just doing an > HTTP GET on the URL. No need for a suffix. However, I personally prefer the > WS-MetadataExchange approach because it fits better with SOAP and its > transport protocol-independent. Also, it allows other metadata information > to be transmitted and I would argue that it's a very simple spec. However, > that's just me. > > Regards, > -- > Savas Parastatidis > http://savas.parastatidis.name > > > > > From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org] On > Behalf Of Cutler, Roger (RogerCutler) > Sent: Thursday, July 01, 2004 10:46 PM > To: Savas Parastatidis; www-ws-arch@w3.org > Subject: RE: Requesting WSDL Files > > Hmmm. Looks like a pretty heavyweight mechanism for such a simple task. > Although you're right that it's not fully general, it seems to me the simple > "?wsdl" HTTP method gets the 80-20 ... and it sure is simple. > > -----Original Message----- > From: Savas Parastatidis [mailto:Savas.Parastatidis@newcastle.ac.uk] > Sent: Thursday, July 01, 2004 4:37 PM > To: Cutler, Roger (RogerCutler); www-ws-arch@w3.org > Subject: RE: Requesting WSDL Files > Dear Roger, > > I don't think that there is a specification and I feel that one would be > unnecessary. The ?WSDL suffix can be used when HTTP is involved but how do > we get the WSDL of a Web Service when we use TCP/IP or SMTP or any other > protocol? That's the reason for the existence of the WS-MetadataExchange > specification. That will be the way to go. If you know the endpoint of a Web > Service, then you can ask it for its WSDL, its policy, etc. > > http://msdn.microsoft.com/webservices/understanding/specs/default.aspx?pull=/library/en-us/dnglobspec/html/ws-metadataexchange.asp > > Regards, > -- > Savas Parastatidis > http://savas.parastatidis.name > > > > > From: www-ws-arch-request@w3.org [mailto:www-ws-arch-request@w3.org] On > Behalf Of Cutler, Roger (RogerCutler) > Sent: Thursday, July 01, 2004 7:27 PM > To: www-ws-arch@w3.org > Subject: Requesting WSDL Files > > Here's a question that is sort of WSA-like. I guess. > We have some experience with WS interop, but so far it's all one direction: > Web service on Windows server, clients on other platforms. Sooner or later > we will want to go the other direction. One really nice feature of the > Microsoft .Net implementation of Web services is that if you append "?WSDL" > (or "?wsdl") to the URL of the Web service it will return the WSDL file. As > far as I know this is not in any spec (I could easily be wrong, of course), > but it's clearly useful and I'm using it. So the obvious questions are: > 1 - Is this indeed part of some spec that I don't know about, so one should > expect it on other platforms? > 2 - If not, have other major vendors been doing this too? Is it by any > stretch becoming a de facto standard? > 3 - If so, is there any case preference on platforms that tend to be more > case sensitive than Windows? >
Received on Friday, 2 July 2004 14:42:11 UTC